Making Basic IT Services Great Again
September 23, 2024 Alex Woodie
The focus of the tech world has been squarely on generative AI since ChatGPT descended the golden staircase and entered our hearts and minds in late 2022. Since then, building GenAI apps and services has become the number one priority for businesses around the globe. Unfortunately, the GenAI fixation has come at the detriment of basic IT services, a recent IBM study suggests.
The level of hype for generative AI the past two years has been off the chart. Not since the initial dot-com boom of the late 1990s has a new technology triggered such a tsunami of interest, some of it good and some of it bad. On the one hand, some prominent deep learning researchers said GenAI was the equivalent of nuclear weapons and would spell the end of humanity. On the other hand, tech bros are envisioning a new world order marked the end of rote office work and more leisure time.
Like most technological innovations, the reality is that we’ll likely end up somewhere in between. Yes, the technology will likely provide real productivity gains for certain human tasks, particularly those that involve working with large amounts of text. McKinsey last year predicted an annual gain of $2.6 trillion to $4.4 trillion due to GenAI, thanks to greater efficiency in customer operations, marketing and sales, software engineering, and R&D. But no, we’re not on the cusp of artificial general intelligence (AGI), that point when machines actually begin to think for themselves. Large language models (LLMs) themselves make fundamental logic errors, can’t actually think, and aren’t sufficient to close the gap with AGI, the New Jersey Innovation Institute says.
With that said, the already demonstrated capabilities of GenAI, plus the predictions of trillions of dollars in economic growth and business efficiency, are more than enough to open up corporate budgets. IDC predicted businesses would spend about $40 billion globally on GenAI (both hardware and software) this year, reaching $151 billion in 2027. Among the tech firms benefiting a lot from the GenAI gold rush are GPU-maker Nvidia, whose stock price has risen 10X from late 2022 and is now worth $2.85 trillion, and ChatGPT-maker OpenAI, which is reportedly close to sealing a massive $6.5 billion funding round (the world’s largest, apparently) at a $150 billion valuation. For comparison, IBM’s market cap is currently $200 billion (thanks to a nice 48 percent bump in Big Blue’s stock valuation over the past 12 months).
2024 CxO Study – 6 blind spots tech leaders must reveal.
With the hyper-focus on GenAI, it shouldn’t come as a surprise that IT departments are getting worse at delivering “basic IT services.” That’s the conclusion of a new IBM Institute for Business Value paper, titled “2024 CxO Study – 6 blind spots tech leaders must reveal.” The 35-page report, which you can download here, is based on a survey of 2,500 C-suite tech leaders done in the first quarter by Oxford Economics.
The survey made some stark conclusions. For starters, the percentage of tech leaders (CIOs, CTOs, CDOs) who say the IT department is “effective at providing basic technology services” has gone from 69 percent in 2013 to 47 percent in 2024. CEOs have an even less rosy view on their IT departments’ competency at delivering core IT services, going from a 64 percent effective rate in 2013 to 36 percent today.
Why the massive decline in core IT competency? There are likely a number of reasons, but one of them undoubtedly is budget allocations. To that end, the figures aren’t likely to turn around any time soon, considering that it found that hybrid cloud and AI spending will go from consuming 41 percent of the IT budget to consuming 50 percent in two years. That means fewer dollars to pay for RPG developers, IBM i administrators, Db2 for i engineers, and security experts to close all the holes new tech creates.
However, rather than questioning the wisdom of moving half of your IT chips to emerging tech or public cloud, IBM uses the slow-motion collapse of core IT competency across the world’s IT departments as a solid rationale for accelerating the transition to GenAI Shangri La.
“IT as a standalone function is dead,” IBM says in its CxO study. “The rapid ascent of generative AI delivered the death knell. Technology is the business. And 72 percent of top-performing CEOs say competitive advantage depends on who has the most advanced generative AI.”
So instead of stopping the bleeding in IT departments that’s hurting their ability to deliver core IT services – and potentially putting the company’s survival at stake – IBM wants tech leaders to double-down on GenAI. It wants them to adopt a “fresh, bold approach to innovation.” All change is good, and since resistance to change represents a “top barrier to innovation,” tech leaders should overcome it by “amping up their outreach” and “evangelizing a culture for innovation.” Whatever you do, don’t worry about “near-term concerns such as efficiency…”
2024 CxO Study – 6 blind spots tech leaders must reveal.
Do you have existing IT systems that are working and getting the business of the business done? That’s great, but instead of resting on your laurels, tech-forward leaders will use it as an excuse to push the company to rip it out in favor of something that’s in some way connected to that digital star in the making, GenAI.
“Refactor legacy systems for AI readiness,” IBM says, “Reframe legacy infrastructure challenges as business impediments preventing rapid gen AI adoption at scale.”
To be sure, there are plenty of problems with aging, out-of-date systems, and lots of room for improvement. The IBM i platform and its Big Iron brother, the System z mainframe, are the epitome of legacy systems. There’s a lot of good code and apps out there running on Power and System z, and a lot of bad code and apps too. For decades, companies have tried to figure out how to modernize their aging RPG and COBOL applications and reshape them to meet changing business requirements. When that fails, they look for ways to move these apps to industry-standard servers or “the cloud.” Some of these “legacy modernization” projects succeeded, and many of them fail spectacularly, due to a variety of reasons we have covered to death here in these IT Jungle pages.
The folks at Big Blue that build IBM i and System z and support those customers are looking for ways to leverage GenAI in ways that actually help them. System z has a watsonx co-pilot for COBOL code modernization, and Rochester is building something similar for RPG. These projects will likely have a positive impact on the IT departments of IBM customers, who are already being strained with flat to declining IT budgets, which will lower their ability to provide basic IT services, as IBM itself documented in the CxO study.
Tech leaders should tread carefully with their GenAI projects, particularly now that the hype around GenAI has peaked and begins to decline. It turns out that getting value out of GenAI is a lot harder than was first envisioned. You can’t just slap some GPT on your data or apps, and expect to see the dollars roll in. There are legal, ethical, and regulatory risks that companies will open themselves up to by rolling out half-baked GenAI products, which is why so few GenAI products have made the leap from development to production. And companies are slowly coming to the realization that their data is in horrible shape and really not sufficient for supporting GenAI.
The GenAI revolution is happening, and it will create separation in the market. Companies that have done the hard work around data governance, building a data-driven culture, and thinking about the ethical side of it have a big advantage over those that haven’t done that work yet. That’s one big reason why companies should be careful not to get carried away with GenAI salesmen who come bearing buzzwords like “reinvention” and “self-disruption.”
Shifting IT dollars from the “basic IT services” column into “speculative GenAI investments” column may seem like a rationale thing to do for someone who truly believes that the GenAI meteorite is about to hit. But the reality is that the GenAI meteorite is a lot further from Earth than some people are making it out to be.
If being good at delivering basic IT services gets any scarcer, then maybe that is what ends up truly separating the business haves from the have-nots.
RELATED STORIES
GenAI Interest ‘Exploding’ for Modernization on IBM i and Z, Kyndryl Says
Getting A Handle On What GenAI Might Cost
Could a CrowdStrike-Type Outage Hit IBM i?
Some Thoughts On Big Blue’s GenAI Strategy For IBM i
How To Contribute To IBM’s GenAI Code Assistant For RPG
Good call to sobriety. The thing is: GenAI in companies requires a very good foundation in core IT data to train on, good data semantic and security model. Random excel files? Not its thing. Well structured database, with relations and properly documented, a datawarehouse with good data architecture and data semantics? Much better to train on, and I see value i.e. in traditional BI product to support even sophisticated queries in natural language respecting data security and user authorizations.
An external lengthy remote model train is tolerable, but the inference part should be done locally to have more control in the performance characteristics (I refer here to ERP data).